Machine learning-based dynamic CEA trajectory and prognosis in gastric cancer

Area of Science:

• Oncology
• Biomarkers
• Cancer Research

Background:

• Carcinoembryonic antigen (CEA) is a known prognostic marker in gastric cancer.
• The prognostic significance of dynamic CEA level changes over time remains under-explored.

Purpose of the Study:

• To investigate the prognostic value of perioperative CEA level trajectories in gastric cancer patients.
• To identify distinct CEA patterns and their association with patient survival outcomes.

Main Methods:

• Analysis of perioperative CEA levels (pre-surgery, early post-surgery, late post-surgery) in 578 gastric cancer patients.
• K-means clustering to define CEA trajectories.
• Kaplan-Meier analysis and Cox regression to assess survival differences.

Main Results:

• Three distinct CEA trajectories (high, medium, low) were identified.
• Higher CEA trajectories correlated with significantly worse disease-free survival (DFS) and overall survival (OS).
• The high CEA trajectory group showed over double the mortality risk compared to the low trajectory group (HR 2.64).

Conclusions:

• Dynamic CEA trajectories are significant independent prognostic factors in gastric cancer.
• Patients with higher CEA trajectories require enhanced monitoring due to poorer prognosis.
• Monitoring CEA trends offers valuable insights beyond static measurements for gastric cancer management.